Recommendations: Upsells, Cross-sells & Recipes

Recommendations are the key to increasing your Average Order Value (AOV) and units per transaction. By showing shoppers the right products at the right time, you facilitate discovery and encourage impulse additions to the cart.

Unlike basic "featured product" blocks, Cavaco AI's recommendation engine is fully dynamic, powered by Recipes—intelligent algorithms that determine exactly which products to display based on context, history, and behavioral data. Furthermore, these recommendations can be embedded not just on your website, but directly within your email and SMS campaigns to create a continuous, omnichannel shopping experience.

Top Ecommerce Use Cases & Placements

To maximize revenue, you should deploy specific types of recommendations at different stages of the customer journey. Here are the most effective strategies:

1. The Homepage Discovery (Trending & New Arrivals)

When a shopper lands on your homepage, they are often browsing without a specific product in mind. Your goal is to show them what's popular or new to capture their interest.

• The Recipe: Use a Recipe built with Trending Items or New Arrival Items ingredients.
• The Personalization: Apply the Affinity Personalization condiment. If a returning shopper has a high affinity for the "Mens" category based on past behavior, the engine will automatically re-rank the trending items to show men's products first.
• Placement: A carousel high up on the homepage.

2. The PDP Cross-Sell (Frequently Bought Together)

When a shopper is viewing a specific product, they are highly engaged. This is the perfect time to offer complementary items (e.g., batteries for a camera, a matching belt for shoes).

• The Recipe: Use the Co-occurrence Items ingredient, specifically tracking the cobought (purchased together) or coviewed events.
• The Rule: Use a Context Exclusion condiment so the engine never recommends the exact product they are currently looking at.
• Placement: An inline grid or small carousel directly below the "Add to Cart" button on the Product Description Page (PDP).

3. The PDP Alternative (Similar Products)

If the shopper isn't quite sold on the product they are viewing (maybe it's too expensive or the wrong color), keep them on your site by showing alternatives.

• The Recipe: Use the Similar Items ingredient, matching on attributes like category, price tier, or material.
• Placement: A carousel at the very bottom of the PDP.

4. The Cart Upsell (Don't Forget These)

The shopper is ready to buy. Offer low-friction, high-margin items that make sense based on what's in their cart.

• The Recipe: Use Co-occurrence (based on cart contents) or Recent Items (reminding them of something they looked at earlier but didn't add).
• The Rule: Use a History Exclusion condiment to ensure you don't recommend something they already purchased in the last 15 days.
• Placement: Inside the slide-out cart drawer or on the dedicated cart page.

Deep Personalization: How Affinities Work

Cavaco AI doesn't just show the same products to everyone. It builds a unique profile for every shopper based on their behavior, calculating Affinities to personalize the experience.

How it's calculated: As a user browses, clicks, and buys, Cavaco tracks the attributes of those products (e.g., Category = "Running Shoes", Brand = "Nike", Color = "Black"). Over time, the engine calculates an affinity score for these attributes.

How to use it: When building a Recipe, you apply the Affinity Personalization condiment. You map a target_attribute (the attribute of the products being recommended) to an affinity_attribute (the attribute the user likes), and assign a weight (0 to 1).

Example: You have a "Best Sellers" recipe. You add an Affinity Condiment mapping the product's "Category" to the user's "Category Affinity" with a weight of 0.8. Now, when a user who mostly buys "Outerwear" views the Best Sellers widget, the engine will pull all best sellers, but mathematically boost the Outerwear items to the first visible slots.

Omnichannel Continuity: Site to Direct Messaging

The power of Cavaco AI is that the recommendation engine isn't limited to your website. You can embed the exact same dynamic Recipes into your Email and SMS campaigns.

This creates omnichannel continuity. If a shopper abandons their cart, the recovery email doesn't just show the abandoned item—it can include a "You Might Also Like" block powered by a Co-occurrence recipe based on that specific abandoned item.

How to set it up:

1. In the Email Template builder, drag the Recommendation block into your layout.
2. Select the Recipe you want to use (e.g., "Trending Personalized").
3. When the email is sent, Cavaco AI calculates the recommendations at the moment of send, using the specific recipient's profile and affinities to populate the products.

By connecting on-site behavior to direct messaging recommendations, you ensure that every touchpoint feels highly relevant and curated to the individual shopper.

Merchandising Controls (Rules)

Even with AI, you sometimes need manual control over merchandising. Widget-level rules let you intervene:

• Boost: Multiply the ranking score of specific products (e.g., push high-margin house brands to the top).
• Bury: Decrease the ranking score of products (e.g., push low-inventory items to the back).
• Pin: Force a specific product into an exact slot (e.g., Slot 1 is always the new flagship product, Slots 2-4 are AI-driven).
• Include/Exclude: Hard rules to force products in or keep them out of the widget entirely.